Use when the user explicitly requests preparing, importing or updating Agent_Engineering_Kit rules in a new or existing project. Ordinary coding, a general audit or missing instruction/model files alone do not trigger it.
Installs into .claude/skills of the current project.
Are you the author of Aek Integrate Project?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/vitaliikhomiakov-aek-integrate-project)
---
name: aek-integrate-project
description: Use when the user explicitly requests preparing, importing or updating Agent_Engineering_Kit rules in a new or existing project. Ordinary coding, a general audit or missing instruction/model files alone do not trigger it.
---
# Integrate project rules
Deliver a reviewable preparation or the authorized import/update. Use the source
framework's `MIGRATION.md` as the procedure owner; preserve the selected stage
checkpoints. This is an instruction-only entry, with no installer; full integration
includes the procedure's required model-selection step through the existing model skills.
Read this entry and the model skills from the identified source by default; their
use does not install them in the target. Source `MIGRATION.md`, "Setup procedures
and persistent skills", owns explicit installation choices and existing-copy handling.
## Resolve the source and target
Reuse the supplied source, target roots/checkouts, client, intent and canonical
task. For an existing adoption, read target `.agents-framework/adoption.md` for
source identity and owned scope. Ask only for material missing information.
When read directly at source `templates/skills/aek-integrate-project/SKILL.md`,
the framework root may be three directories above this skill folder only if both
`MIGRATION.md` and `standards/catalog.toml` verify that source context. An inactive
adopted copy is not the full source library. From an adopted bundle or native
installation, resolve the accessible
framework source from the request or adoption record, independently of the skill's
directory. Never assume the target contains MIGRATION or is the source library.
If that source is unavailable, identify the missing resource before dependent
preparation/application; ordinary work continues through reachable local rules.
Do not invent policy or globally install another copy.
## Read the relevant procedure
Paths in this table resolve from the identified framework source. Select sections
by their headings; follow further references only for the requested operation.
| Work | Read |
| --- | --- |
| Any import/preparation | MIGRATION: Import inputs and boundaries; Prepare and apply; Execution rule for the migration |
| New/existing target and conflict inventory | MIGRATION: New and existing projects; Phase 1; Resolving active conflicts; relevant Phase 2 profile choices |
| Bundle and target entries | MIGRATION: Setup procedures and persistent skills; Layout and path contract; `standards/catalog.md`, selected catalog profiles and relevant Phase 3 instruction restructuring |
| Selected project task skills | MIGRATION: Optional project task skills; Adapting installation routes; selected templates and native mappings in the adoption record |
| Model selection/reuse | MIGRATION: Model-selection step; `aek-model-setup` or `aek-model-reassign` as selected there; source `standards/model-configuration.md` owns actual settings |
| Proposed/applied record | MIGRATION: Adoption record and ownership; `templates/framework/adoption-record.md` |
| Repeat/update or selected relocation | MIGRATION: Repeat and update; Relocating an existing import; current record and actual managed target contents |
| Before writes or interrupted work | MIGRATION: Partial application and recovery |
Prepare source/profile/path/action/content and conflict decisions under that
procedure. Preparation-only intent leaves the target untouched. For application,
reuse existing authorization, preserve originals and reconcile intervening edits;
replace loaded conflicts while retaining required local contracts and gates.
Keep existing task stores and model choices. Pass adoption intent, selected mode,
prior choices, source/target roots and write scope to the model procedure. Reuse
consistent selections and explicit deferrals without repeated questions; report
unresolved setup separately. If a model skill is unavailable, use its reachable
contract and disclose the fallback; an inaccessible contract blocks only dependent
setup. Preparation remains write-free. Preserve scope, typing and proportionate verification
rules when adapting entries; catalog availability is not unconditional reading.
Report prepared/applied/unchanged/partial scope, evidence, unresolved decisions and
the next agreed checkpoint, including the model-step outcome and its owner references.
Distinguish valid files and saved choices from observed client discovery/activation.
Neither source-based reading nor copying this skill proves native activation.
For task skills, report the selected native subset separately from available
bundle templates. Include those native paths in comparison, before-write protection
and explicit installed-file checks; a same-name local skill is not owned by default.